An attempt to crack the melanoma-treatment nut has researchers working on the means for personalized cures through automated literature search. Their goal is to develop a targeted therapy database that matches patient molecular profile with current cancer biology knowledge.
The literature search targets scientific evidence about three types of molecular interactions: those involving the anti-melanoma efficacy of a therapeutic agent; those involving the toxicity of any such agent; and those involving an increase or decrease in efficacy following modulation of functional state. Data in the database are annotated manually and then gathered in a formal representation for computational analysis. Algorithms identify prevalent therapeutic hypotheses and rank treatments based on patient molecular profile.
The database is publicly available. Validation of prediction performance remains before the model can be implemented in the clinic.
- here's the paper